Community-driven content moderation. Voting on scams, copyright, and inappropriate content with micro-rewards for valid votes and escalating strike penalties.
Content moderation is decentralized -- the community votes on what to allow or remove, with token rewards for valid moderation votes.
| Action | Target Types | Purpose |
|---|---|---|
DISABLE_SCAM | Listing, Collection, Profile | Hide malicious or deceptive content |
COPYRIGHT_STRIKE | Listing, Collection | Content violates intellectual property |
INAPPROPRIATE_CONTENT | Listing, Collection, Profile | NSFW, illegal, or policy-violating content |
RESTORE | Listing, Collection, Profile | Reverse a previous moderation action |
OTHER | Any | Custom moderation reason |
- Each voter can only vote once per (targetId, action) combination
- Votes include a text reason explaining the voter's position
- AI-assisted votes are flagged: llmAssisted, llmModel, llmConfidence
- resolved = true, action is executed (hide listing, flag profile, etc.)
- All voters on the winning side earn MODERATION_REWARD token claims
moderationOutcomes tracks running approveCount and rejectCountValid moderation votes earn token rewards:
MODERATION_REWARDNodes that violate protocol rules accumulate strikes with escalating penalties:
| Strike Count | Penalty |
|---|---|
| 1 | Warning |
| 2 | 24-hour fee suspension |
| 3 | 72-hour fee suspension |
| 4+ | Permanent ban (bond slashed) |
Strikes decay over time -- good behavior reduces your strike count. The strike system is separate from user-facing content moderation; it targets node operators who submit invalid attestations or fail PoA challenges repeatedly.
During beta, the foundation master node (Node-1) retains emergency moderation powers. All foundation moderation actions are: